[Soot-list] Regarding Call Graph Construction.....

[Saswat Anand]

It is not clear whether you want to parallelize the points-to analysis
(i.e., Spark) or just the call-graph construction steps. I suppose you
want to do the first one. If so, an orthogonal approach to doing so is
to parallelize the BDD engine "buddy" that Paddle uses for points-to
analysis. I am sure this approach is also challenging. However, the
advantage of this approach is that you do not have to tackle the large
code base of soot, and buddy code is small and well-documented. There
is some evidence that buddy can be significantly more efficient on
multicores using Cilk++ framework.

Saswat